GoodFuture International Algorithm Team Wins Champion and Runner‑up in the 5th Educational Data Mining Workshop

The 5th Educational Data Mining in Computer Science Education Workshop announced its second‑stage results, with the GoodFuture International Algorithm Team and Jinan University Guangdong Smart Education Research Institute standing out among 95 participating teams, earning a champion in one task and a runner‑up in another.

The workshop, co‑organized by North Carolina State University, University of Pittsburgh, Carnegie Mellon University and University of Helsinki, attracted numerous domestic and international universities and research institutes.

Focusing on the "intelligent assessment" problem in computer education, the competition required teams to predict future student performance using past programming problem interactions, aiming to enable automated, personalized evaluation.

For Task 1, GoodFuture adapted cutting‑edge recommendation algorithms, reframing the problem as a recommendation task and modeling student interaction behavior, which allowed them to predict multiple future problems and achieve the runner‑up position.

Task 2 demanded prediction of final semester grades from half‑semester performance; GoodFuture extracted key features and employed a multi‑stage approach to enhance model stability, ultimately securing the champion award.

Their technology, previously a winner of the NeurIPS 2020 Education Challenge, originates from an internal "course‑drop prediction" system that analyzes interaction data to forecast student behavior.

GoodFuture’s AI‑education research has earned international recognition, with papers accepted at the International Conference on Artificial Intelligence in Education (AIED) and citations by the EU, UNICEF, UNESCO, and inclusion in global "China solution" reports.

The team plans to continue deepening research in educational technology, open relevant datasets and technical results, and promote the further development of intelligent education worldwide.